Official Review · ICLR 2025

The authors of this paper explored a compositional workload model for embodied multi-agent cooperation. Unlike previous approaches to embodied multi-agent cooperation, they introduced an explicit world model that simulates the next state by incorporating the estimated actions from the Intent Tracker along with the agent's actions. This model stands out from other single-agent, model-based reinforcement learning approaches by explicitly accounting for multi-agent cooperation dynamics. Utilizing this world model, they implemented a tree search planning strategy similar to Monte-Carlo Tree Search [1]. Their approach demonstrated superior performance in embodied multi-agent cooperation tasks and, due to the compositional nature of their model, exhibited strong generalization capabilities. [1] Silver, David, et al. "Mastering the game of Go with deep neural networks and tree search." nature 529.7587 (2016): 484-489. - They designed explicit compositional world model. It is one of the most distinguishable designs in their modeling, which supports the look ahead planning (tree search planning), and they showed it is beneficial through the empirical evaluation results. As a part of this, the world state estimation is good to build the world model for multi-agent setting. - It is a well written paper. I can easily follow their discussions without unnecessary questions. - The room of the evaluated benchmarks is too small to show the effectiveness of their proposed modeling. In Table 1, COMBO outperformed previous works. Although, when comparing with LLaVA, it shows comparable success rate except TDW-Cook Cooperator 1 setting. Their efficiency on solving the tasks is clearly better than LLaVA, but we felt it is not good enough to evaluate the effectiveness of their modeling. - The generalization performance evaluation is too weak to show that in lines 521-524 and Table 5. They trained the model with 4 agents and applying it to 3 agent only, then what happens if we test with 2 or much more agents such as 10? Additionally, COMBO is more efficient than LLaVA, but LLaVA also showed good success rates for 3 agents setting. The LLaVA is trained on the 3 agents setting? If yes, it should be mentioned in the caption of the figure or the paragraph. If no, then we think this empirical evidence is weak to show the generalizationality of the COMBO. - In Figure 1, what means the surrounding figures? Maybe we guess it is to visualize the look ahead planning, but I am not sure it is a good visualization for showing that.